Comparison of Myocardial Tagging and Feature Tracking in Patients with Severe Aortic Stenosis

Background and aim of the study: Transcatheter aortic valve implantation (TAVI) has become an effective treatment for severe aortic stenosis (AS) in patients considered to be at high surgical risk. However, data relating to myocardial function and deformation in these patients are rare. Cardiovascular magnetic resonance (CMR) tagging has been established as non-invasive technique for the accurate measurement of myocardial motion. However, additional tagging datasets are necessary and the post-processing procedure is time-consuming. Recently, the novel technique of feature tracking (FT) was introduced, with which myocardial strain can be derived directly from balanced steady-state free precession (bSSFP) cine sequences. The study aim was to compare tagging with FT in patients with high-grade AS and who had been considered for TAVI. Methods: Thirty patients with severe AS underwent cardiac magnetic resonance imaging at 1.5 T (Philips Achieva). A stack of serial short-axis slices was used to assess left ventricular end-diastolic volume (LVEDV), left ventricular end-systolic volume (LVESV), and LV mass and function. Tom.Tec 2D Cardiac Performance Analysis software was used to derive myocardial deformation parameters from three short-axis cine views (basal, medial and apical levels) using bSSFP. All patients also underwent myocardial tagging on identical short-axis views, and data acquired from FT and tagging were compared. Results: The mean left ventricular ejection fraction (LVEF), LVEDV and LVESV (57 +/- 14%, 158 +/- 58 ml, and 74 +/- 48 ml, respectively) were all normal. All patients demonstrated concentric hypertrophy of the left ventricle with an increased basal septal thickness (16 +/- 3 mm), LV mass (126 +/- 40 g) and indexed LV mass (68 +/- 23 g/m(2), papillary muscles excluded). Bland-Altman analysis revealed a reasonable agreement between basal and medial circumferential strain (cc), and a good correlation was observed between tagging and FT for the derived basal and medial cc (Pearson's correlation coefficient 0.83 for basal, 0.74 for medial). Basal rotation was impaired compared to previous studies in patients with AS, but medial and apical rotations were comparable. The apical peak diastolic rotation velocity was faster compared to earlier analyses. Rotation, peak systolic and end-diastolic rotation velocity did not demonstrate any correlation or acceptable agreement between FT and tagging. The inter-observer agreement as assessed by the intraclass correlation coefficient for FT showed good results for the cc (basal 0.94, medial 0.83, apical 0.75). Conclusion: To the present authors' knowledge, this is the first study to assess myocardial motion using tagging and FT in patients with AS prior to TAVI. A reasonable agreement was found for the basal and medial cc between both techniques. Estimated cc-values using FT were systematically higher than those with tagging, but this might have been due to the different techniques employed. Although FT is a promising method to assess cc, further studies - including the evaluation of standard values - and investigations with different diseases are necessary.